Ultrasonic-assisted magnetic abrasive finishing (UAMAF) process has been developed to reduce the finishing time of hard materials. In this process, in addition to the rotation of the magnetic tool, ultrasonic vibrations are also applied to the workpiece simultaneously in the longitudinal direction. In the present work, the effects of UAMAF process parameters, including ultrasonic power, the rotational speed of the magnetic tool, working gap, and weight of magnetic abrasive particles (MAPs) on the percentage change in surface roughness (%ΔRa) of 1.2738 tool steel are discussed with an experimental approach. In this regard, besides the development of the UAMAF process, the sintering of MAPs with glass powder is proposed. The experiments are designed according to the Taguchi method. Then, the input parameters of the UAMAF process are modeled and optimized using the signal-to-noise (S/N) ratio analysis to achieve the maximum %ΔRa. Statistical analysis of experimental data shows that the most significant contribution in improving %ΔRa is attributed to the weight of MAPs. Furthermore, according to the comparative study, %ΔRa in the UAMAF process for finishing the DIN 1.2738 tool steel is improved by 86.62% under optimized conditions. In comparison, this value in the MAF process is equal to 48.62% under the same conditions. The results of the surface morphology also underline that abrasive particles in the UAMAF process hit the peaks of the surface roughness due to vibration applied to the workpiece, which leads to a high-quality finished surface.